TMCnet Feature
October 22, 2021

Why automated cloud cost optimization is becoming the new normal

Keeping cloud expenses under control is a challenging job when scaling cloud resources is so simple. Businesses often accept the cloud spend increase as a price to pay for the cloud’s scalability and flexibility. And engineers have grown accustomed to overprovisioning, wasting even 30% of resources.

Is there any hope? Automatic cloud cost optimization comes to the rescue. Continue reading to understand how automation is already helping businesses to reduce their cloud costs and increase their profit margins without any extra work for engineers.

4 approaches to managing and optimizing cloud costs

1. Cost visibility and monitoring

Companies use cost allocation, monitoring, and reporting tools to figure out where the expenses are coming from. Real-time cost monitoring is especially useful since it sends real-time alerts when cloud costs begin to spiral out of control. 

An alert can save you hundreds, if not thousands, of dollars. A computing operation left running on Azure resulted in an unexpected cloud charge of over 500k for one of Adobe (News - Alert)'s teams. One single warning would be enough to avoid that.

Cost budgeting and forecasting 

You can estimate how many cloud resources you'll need and plan your budget - provided that you have enough historical data as a foundation and a clear idea of your future requirements. 

Even cloud-native tech giants struggle to forecast their costs. So many people used Pinterest over one holiday season that the company's bill went much above its initial estimates and Pinterest had to purchase additional capacity at a much higher price tier.

Legacy cost optimization 

Traditional cost optimization tools compile all of the information listed above to create a detailed picture of your cloud spending and discover possible candidates for optimization. There are a number of technologies that can assist with this - but they usually just provide static recommendations for engineers to execute. And the time required to implement them reduces the impact of optimization on your gross margin.

Cloud native cost optimization 

Optimizing cloud expenses is often a point-in-time task that requires both effort and skill. Striking a balance between cost and performance is hard. But modern cost optimization tools use automation to address changes in resource demand or supplier price, allowing for significant cost reductions.

Should we continue to rely on engineers to do all the optimization tasks by hand? Automation is a viable alternative, here’s why.

How does the automated approach to optimization work?

Ask any team dealing with cloud expenses and you’ll hear that they need a lot of time to allocate, comprehend, analyze, and anticipate these costs. And this is just the beginning. Engineers still have to make infrastructure improvements, study pricing plans, spin up additional instances, and more.

Automation relieves teams from that duty. At the same time, teams always have full control over what occurs - they just don't have to do anything proactively to save money.

An automated cloud cost optimization solution:

  • Selects the most cost-effective instance types and sizes to meet the applications' needs.
  • Autoscales cloud resources to cope with demand spikes.
  • Removes resources that haven't been used in a while to avoid cloud waste.
  • Makes use of spot instances and gracefully manages unexpected disruptions.
  • It automates storage and backups, security and compliance management, and changes to configurations and settings to help teams save money in other areas.
  • It implements all of these modifications in real time, mastering cloud optimization's point-in-time nature.

Here’s an example of automated optimization

Imagine a combination of AWS On-Demand and spot instances used to operate an application. By using cloud automation platforms, you can evaluate the configuration and find the most cost-effective spot instances. Let’s say that you require a system with 8 CPUs and 16 GB of RAM (News - Alert).

So you run the analysis and the solutions assigns the workload to an instance called INF1, which has a strong ML-specialized GPU. It's a supercomputer, and they're generally pretty pricey.

Why did the platform choose this particular instance? Take a closer look at the cost and you’ll discover that INF1 was actually cheaper than the general-purpose compute you were using at the time. You’d probably never have thought to check for spot instances in this category, simply assuming that it’s more expensive. As a result, you could have lost out on this hidden gem.

If you're still not convinced that automation is the best way to control enterprise cloud costs, here are a few additional reasons why a manual configuration is no longer an option.

3 reasons why we should say goodbye to manual cost optimization

1. Humans minds find cloud billing hard to understand

Cloud bills are long, complicated, and hard to understand. Each service has a defined billing metric for it, so understanding your team’s usage or making confident predictions about it is next to impossible. And what if several teams or departments contribute to one bill? Then you’re looking at a serious cost allocation challenge. 

2. Forecasting is guesswork for many

To forecast your future resource demands, teams need to do a lot things:

  1. Analyze usage reports to learn about spend patterns,
  2. Identify peak resource usage scenarios by looking at periodic analytics and running reports over usage data,
  3. Consider other sources of data (for example, seasonal customer demand patterns). If they correlate with peak resource usage, you might be able to identify them before they happen,
  4. Monitor resource usage reports and set up alerts,
  5. Measure application- or workload-specific costs,
  6. Calculate the total cost of ownership of the cloud infrastructure, 
  7. Analyze the pricing models of cloud providers you’re using, 
  8. Plan capacity requirements over time, 
  9. Aggregate all of this data in one place to understand your costs better.

Looks like a lot of work, doesn’t it? And many of these tasks aren’t one-off jobs, but activities teams need to do regularly. 

3. Manual instance rightsizing doesn’t make sense

AWS has almost 400 different instances. Analyzing them all manually would take a lot of time.


  • To pick the best instance for the application, teams need to carry out these tasks manually: Defining minimum requirements across compute dimensions including CPU (architecture, count, and the choice of processor), Memory, SSD, and network connectivity. 
  • Selecting the right instance type,
  • Choosing the instance size that can handle your scale,
  • Analyzing different pricing models (On-Demand, Reserved Instances, Savings Plans, spot instances, and Dedicated Hosts).

Automation is the new normal

Scaling, deploying, and configuring cloud resources manually may lead to errors that impact an application’s availability or performance. 

By eliminating human work in managing and optimizing cloud resources, companies can accelerate their processes and spend less time on diagnosing and debugging their infrastructures. Most importantly, automation brings cost savings that are both immediate and guaranteed.

» More TMCnet Feature Articles


» More TMCnet Feature Articles